Creative Education, 2018, 9, 615-624

http://www.scirp.org/journal/ce
ISSN Online: 2151-4771
ISSN Print: 2151-4755

Using Collaborative Agent-Based Computer

Modeling to Explore Tri-Trophic Cascades
with Elementary School Science Students

Anthony J. Petrosino1, Max K. Sherard1, Jason R. Harron1, Walter Stroup2

Department of Curriculum and Instruction, University of Texas at Austin, Austin, USA

1

University of Massachusetts-Dartmouth, Dartmouth, USA

2

How to cite this paper: Petrosino, A. J., Abstract

Sherard, M. K., Harron, J. R., & Stroup, W.
(2018). Using Collaborative Agent-Based This paper investigates an in-service teacher and her student’s abilities to util-
Computer Modeling to Explore Tri-Trophic ize, implement, and enact a participatory agent-based modeling program, de-
Cascades with Elementary School Science
veloped as part of the group-based cloud computing (GbCC) for STEM Edu-
Students. Creative Education, 9, 615-624.
https://doi.org/10.4236/ce.2018.94043
cation Project funded by the National Science Foundation. In this first cycle of
design-based implementation research with an in-service teacher and her 300
Received: February 27, 2018 students, we examine student participatory learning and teacher experience.
Accepted: April 15, 2018
By implementing models with teachers, we intend to 1) improve iteratively
Published: April 18, 2018
the GbCC learning technologies and 2) develop more informed and aligned
Copyright © 2018 by authors and pedagogies for teaching in socially mediated and generative learning envi-
Scientific Research Publishing Inc. ronments.
This work is licensed under the Creative
Commons Attribution International
License (CC BY 4.0).
Keywords
http://creativecommons.org/licenses/by/4.0/ Group-Based Cloud Computing, NetLogo, In-Service, Modeling
Open Access

1. Introduction
The most conspicuous feature of school-based learning is that it occurs in group
settings; however, instruction typically engages only the individual by following
the initiation-response-feedback sequence (Wells, 1993). For classrooms to en-
gage in more socially mediated and generative teaching and learning, instructors
must design situations for groups of students to construct relations between
stored knowledge, experience, and new information (Wittrock, 1991). Using the
taxonomy of generative design provided by Stroup, Ares, & Hurford (2004),
scientific modeling allows learners to engage in generative learning with the help
of agent-based models deployed using the group-based cloud computing

DOI: 10.4236/ce.2018.94043 Apr. 18, 2018 615 Creative Education

A. J. Petrosino et al.

(GbCC) web-based platform (Petrosino & Stroup, 2017; Stroup, Ares, Lesh, &
Hurford, 2007).
By merging NetLogo Web (Wilensky, 1997) with other open science and ma-
thematic libraries within a highly interactive, browser-based, cloud-supported
architecture, GbCC significantly extends the capabilities of the HubNet system
(Wilensky & Stroup, 2002) for implementing group-based participatory simula-
tions. The result is an environment that allows learners to work collaboratively
to participate in, author, and then share simulations and models of a wide array
of phenomena. The GbCC architecture can be used to develop new environ-
ments and activities or resituate existing models or participatory simulations.
The capabilities support learning across a range of domain content including
disease transmission, population dynamics, physical phenomena, and social stra-
tification. This study illustrates resituating and updating of the Wolves-Sheep
Predation model (Wilensky, 1997; Wilensky & Reisman, 2006), shown in Figure
1.
Rather than function as an individual-user simulation, we could use the GbCC
architecture to create a network-mediated, group-based modeling activity (Pe-
trosino & Stroup, 2017), the Yellowstone Problem, for exploring a specific kind
of complex food web—tri-trophic cascades—where the presence of predators
and producers exerts significant control over population dynamics. We use the
model to explore the consequences of the extirpation of wolves in Yellowstone
National Park and the resulting ecological effects related to the power of
top-down ecosystem control in tri-trophic cascades (Ripple & Beschta, 2012). To
introduce the activity, a short video produced by Sustainable Human (How
Wolves Change Rivers, 2014) was used to illustrate how the reintroduction of
wolves in Yellowstone had a number of impacts on the environment, including
changing the flow of rivers. This scenario is more complicated than typical
science curricula can model with representations like symbolic food webs. The
complexity involved in how wolves shape rivers—what we called the Yellow-
stone Problem—can be modeled using collaborative agent-based models. A
concept map was developed to illustrate some of the agents involved in this
tri-trophic cascade (Figure 2).
The GbCC platform is a powerful computational tool; however, “powerful
technological tools, in the absence of powerful pedagogy, detract from rather
than contribute to learning” (Philip & Garcia, 2013: p. 313). With this insight in
mind, the goal of this paper is to 1) examine how the integrated GbCC environ-
ment informs the learning of ecology concepts and modeling knowledge through
socially mediated forms of classroom activity and 2) determine changes to the
model or activity design for future implementation.

2. Methods
Consistent with the goals of characterizing student participation while building
in-service teacher capacity for pursuing more fully socially mediated approaches

DOI: 10.4236/ce.2018.94043 616 Creative Education

 A. J. Petrosino et al.

Figure 1. NetLogo wolf-sheep predation model. From netlogo wolf sheep predation
model, by Wilensky, 1997, evanston, il: northwestern university, center for
connected learning and computer-based modeling. used via a creative commons
license. this model was used in the 4-day unit to teach students about the
yellowstone problem.

Figure 2. A concept map developed to illustrate the organisms and

ecological processes involved in determining river movement and
health in Yellowstone National Park.

to classroom-based learning, this research represents one cycle of design-based

implementation research (Fishman, Penuel, Allen, Cheng, & Sabelli, 2013). Si-
tuated in a southern state whose standards are based on Next Generation Science
Standards, students are required to learn about food webs, species interactions,
and the negative impact of human actions (Next Generation Science Standards
Lead States, 2013). The standards, however, fail to capture the emergent and

DOI: 10.4236/ce.2018.94043 617 Creative Education

A. J. Petrosino et al.

complex nature of ecosystem dynamics represented in the Yellowstone Problem.

To address this failure, a 4-day ecology unit was generated with teacher input.
This unitcontained: 1) a teacher lesson plan; 2) student materials; 3) resource
packets; and 4) instructions for accessing and using the GbCC Wolf-Sheep Pre-
dation model (Wilensky, 1997). The 4-day unit is described in Table 1.
One science teacher who instructs four fifth-grade and four sixth-grade class-
rooms volunteered to implement the unit after being approached by one of the
authors. Upwards of 300 students were able to explore the complex story of how
the presence of wolves generates a cascade of effects that influence plant and
river patterns in Yellowstone National Park; however, due to incomplete collec-
tion, only 112 student material packets were collected. The student body is clas-
sified by the state as majority economically disadvantaged (62.8%), and the eth-
nic or racial demographics are provided in Table 2.

Table 1. Description of the 4-day unit.

Lesson Description

Students begin the unit by watching the short video created by

Sustainable Man, which describes the Yellowstone Problem (How
Wolves Change Rivers, 2014)a. Students read and analyzegraphics,
1) Introduction to How
visualizations, and other information sources about the agents
Wolves Shape Rivers
involved: wolves, elk, and aspen trees. Students culminate the lesson
by writing a position statement regarding the following prompt:
Should wolves be protected from hunting in Yellowstone?

2) Introduction to the Students explore the GbCC Wolf-Sheep Predation model in pairs
Group-based Cloud with Google Chrome books. Students are given the challenge to run
Computing (GbCC) the model with varying amounts of wolves to note the outcomes.
Wolf-Sheep Predation Students end the lesson by thinking about what makes a strong
Model argument.

Students generate a final position and supporting evidence for or

3) Generating an Argument against protecting wolves in Yellowstone. Students use the GbCC
models to generate evidence to support their position.

Students present their poster displaying their argument for

4) Presentations
protecting wolves in Yellowstone.

a
“How Wolves Change Rivers,” 2014, Retrieved from
https://sustainablehuman.tv/remix/how-wolves-change-rivers

Table 2. Student ethnicity percentages 2015-2016 reported from the state’s department of
education website.

Ethnicity %

Hispanic or Latino 15.5

White or European American 16.8

Black or African American 66.3

The student body is classified as majority economically disadvantaged (62.8%).

DOI: 10.4236/ce.2018.94043 618 Creative Education

 A. J. Petrosino et al.

Student material packets were collected, digitized, and analyzed. These re-
sources allowed us to evaluate the degree to which GbCC capabilities expand
participatory learning. The cooperating teacher offered feedback regarding the
logistics and experiences of using GbCC in the classroom. Findings assisted us in
understanding how well and under what conditions the GbCC model-based les-
sons were successful at expanding participation, engaging students in mod-
el-based reasoning, and uncovering the complex relationships involved in the
Yellowstone tri-trophic cascade (Collins, 1990).

3. Features of the GbCC Wolves-Sheep Predation Model

The GbCC Wolves-Sheep Predation model can be altered or even authored mul-
tiple ways. A list of ways the student can manipulate the model is presented be-
low, along with a screenshot of the full model (Figure 3). In this model the stu-
dents can:
 individually manipulate all sliders and toggles, set up the model, and run the
model by pressing “go”;
 attempt to find values that either cause extinction of the wolves or sheep,
cause overgrazing of the grass, or create an equilibrium between all three
species;
 share their code to the gallery by pressing the share button, which sends their
graph to the classroom gallery space;
 choose a graph from the gallery by pressing the graph button, which imports
the slider values from the chosen graph to the user’s slider values so the user
can run the same model;
 the command center to make changes to the code that runs the model; and
open and author the NetLogo code resulting in direct changes to the func-
tioning of the model.

4. Results and Discussion

The 4-day unit took place over the course of 2 weeks in the teacher’s classrooms.
Students were tasked with generating an argument for or against protection of
wolves in Yellowstone supported with evidence from their GbCC modeling ex-
perience. Specific questions from the student material packet were identified
based on their relevance to the research questions, digitized, and analyzed.
Teacher feedbacks in the form of typed responses are provided at the end of the
section to frame changes made to the model and the activity.

4.1. Day 2, Question 2

Following introduction to the Yellowstone problem, the agents involved, and
background information on the environment, students began Day 2 by directly
engaging with the GbCC Wolf-Sheep Predation model. After learning how to
manipulate sliders and toggles, run the model, and read the graphs, students
were prompted to answer the following question:

DOI: 10.4236/ce.2018.94043 619 Creative Education

A. J. Petrosino et al.

Figure 3. Group-based cloud computing Wolf-Sheep Predation model with sliders, toggles, visualization, graphs, and netLogo
code.

Why would we, as ecologists, use this computer model to learn about wolves,
elk, and aspen—and make our argument to the public?
Students worked in pairs to generate answers to this question. Of the 112 stu-
dent materials packets collected, only 64 contained the page with this question.
The teacher mentioned misplacing the work of the other students, which ex-
plains the missing data. The list was transcribed, compiled, and read. Schwarz et
al. (2009) provided a table of the “candidate components of metamodeling
knowledge for a learning progression for modeling” (Table 1, p. 636). The table
provided three components of metamodeling knowledge: 1) the nature of mod-
els, 2) the purpose of models, and 3) the criteria for evaluating and revising
models. Question 2 elicited student ideas about the component of purposes of
modeling. Student responses were categorized and examples chosen, as shown in
Table 3.
By engaging with the computer model, students were able to easily generate
numerous reasons why this tool could support an effective argument. Most fre-
quently, students mentioned the ease of using the computer model due to the
long-term and large-scale nature of learning about wolves in Yellowstone. Stu-
dents also touched upon ideas of using the computer model to determine which
amount of wolves would be healthy for the ecosystem, which is evidence of stu-
dents thinking about models as generative or predictive tools. Going forward,
activities designed around using GbCC modeling capabilities should involve
more time spent with the models as well as group conversations about the pur-
poses, uses, limitations, and affordances of modeling.

DOI: 10.4236/ce.2018.94043 620 Creative Education

 A. J. Petrosino et al.

Table 3. Example student responses for each of the four subthemes within the purposes
of modeling component.

Component Subcomponent Example student responses

“To see the effect”

“Easier to see”
Models are sense-making tools for
“Our school isn’t in Yellowstone park.”
constructing knowledge.
“Easier on computers.”
“To see how it would work.”

“Because you see the results before doing

Models are communication tools for
it.”
conveying understanding or knowledge.
Purposes of “For training.”
modelinga
Models can be used to develop new
understandings, by predicting new aspects “Learn which numbers work best.”
of a phenomenon

“It can show us and not hurt the

Models are used to illustrate, explain, and environment.”
predict phenomena. “We use simulation because we will know
what [is] going to happen.”

4.2. Day 2, Questions 3 and 4

To stimulate students’ exploration of the model, two questions were provided
that allowed students to see the extremes of the Yellowstone Problem: complete
extirpation of wolves and overpopulation of wolves. The question below allowed
students to simulate the scenarios and take notes on the outcomes:
What happens when you put no wolves in the model? What happens when
you put a lot of wolves in the model?
Students overwhelmingly responded to these questions in the manner that the
researchers and teacher had hoped. For Question 3, of the 112 responses col-
lected from students, 99 responded with some acknowledgement that removal of
wolves would lead to an increased sheep population. Student responses ranged
in detail and sophistication. Some mentioned simply that there were more sheep,
and some generated a more detailed cause-effect response. For example, one
student replied, “The population of sheep increased and there is less grass, then
the sheep decreased.” This student acknowledged the cyclical nature of sheep
(elk) and grass (aspen) populations when left unchecked by the presence of
wolves.
Similarly, when students simulated the model with a large number of wolves,
the majority of students (95 of 112) responded with some answer related to the
declining number of sheep, but relatively few students capture the cyclical nature
of populations. One student responded, “The sheep will decrease because the
wolves ate them all, and also the wolves will decrease because they ate all the
sheep and the wolves will have nothing to eat.”
A possible explanation of the breadth of responses yet lack of depth of student
answers could be related to the amount of time spent allowing students to ex-
plore the models and write. In this 50-minute lesson, students only received

DOI: 10.4236/ce.2018.94043 621 Creative Education

A. J. Petrosino et al.

about 20 minutes of time to explore the model and write about their conclu-
sions. In the future, more time will be allocated for free exploration of the mod-
els.

4.3. Teacher Feedback

The cooperating teacher provided feedback after the series of lessons about the
benefits and challenges of using GbCC in the classroom as well as the nature of
participation and engagement. The responses are listed in Table 4.
The most salient feature of GbCC is the ability to share models to the gallery
space for other students to use; however, the teacher noted that only the sixth
graders used this feature. More research is needed to determine what factors
contribute to students sharing their models to the gallery space and using mod-
els from the gallery space.
Moving forward, this research allows for the continued exploration of (a)
next-generation, fully authorable, collaborative, cloud-based computing focused
on science, technology, engineering, and mathematics (STEM) and (b) partici-
patory approaches to modeling and classroom-based inquiry aimed at advancing
student abilities in STEM-related coursework and careers. Our research
represents one step in the cycle of development and implementation across
many schools, subject areas, and grade levels. Findings suggest that teachers and
students would prefer more time spent with the simulation as well as tasks that

Table 4. Feedback provided from the cooperating teacher.

Question Response

What did you find beneficial
about teaching using
group-based cloud
computing (GbCC),
specifically the wolves and
sheep simulation?
The kids loved it! At the start they understood how serious the loss
of gray wolves was to Yellowstone. They understood even more
when I showed them how the gray wolves population was
extremely important to Yellowstone because it kept Yellowstone in
balance. When they were able to see the simulation of the wolves
and sheep and how the sheep overtook the area and then ate all the
grass, they understood that the need for wolves in order for
Yellowstone to be healthy was every important.

Mostly that the site kept crashing. Some of the buttons and
numbers seemed confusing to the kids, and they didn’t know what
to do with them. Most of the lesson were too fun to find
What did you find challeng-
challenging. For fifth grade, though, I hyped it up and made it my
ing?
own because the lesson wasn’t as “kid friendly” as I hoped it would
be. Getting to discover different species of wolves is something
added to make it more fun.

Did your students use the My fifth graders didn’t, but my sixth graders did. They shared with
Gallery feature of the each other how to keep the area balanced and what process they
simulation? used to do that.

How was participation in

100% participation. They loved getting on the computer and being
your class when using the
able to push buttons and see what happened. Sixth grade liked the
GbCC Wolves-Sheep
gallery feature a lot and shared with people outside of their groups.
Predation model?

DOI: 10.4236/ce.2018.94043 622 Creative Education

 A. J. Petrosino et al.

facilitate students’ use of the gallery share and authoring features. This scalable
and low-cost technological and pedagogical infrastructure is intended to directly
increase student and teacher involvement with high-quality STEM learning ex-
periences.

Acknowledgements
The authors would like to thank the students and teachers who participated in
this study. This material is based upon work supported by the National Science
Foundation under Grant No. DRL-1615207. The opinions expressed here are
those of the authors above.

References
Collins, A. (1990). Toward a Design Science of Education. Technical Report No. 1. New
York, NY: Center for Technology in Education.
Fishman, B. J., Penuel, W. R., Allen, A.-R., Cheng, B. H., & Sabelli, N. (2013). De-
sign-Based Implementation Research: An Emerging Model for Transforming the Rela-
tionship of Research and Practice. Yearbook of the National Society for the Study of
Education, 112, 136-156.
How Wolves Change Rivers (2014).
https://sustainablehuman.tv/remix/how-wolves-change-rivers
Next Generation Science Standards Lead States (2013). Next Generation Science Stan-
dards: For States, by States. Washington DC: The National Academies Press.
Petrosino, A. J., & Stroup, W. M. (2017). Group-Based Cloud Computing for STEM
Education. Columbus: Poster at the Annual Convention of the American Society of
Engineering Education.
https://www.asee.org/public/conferences/78/papers/19018/view
Philip, T., & Garcia, A. (2013). The Importance of Still Teaching the iGeneration: New
Technologies and the Centrality of Pedagogy. Harvard Educational Review, 83,
300-319. https://doi.org/10.17763/haer.83.2.w221368g1554u158
Ripple, W. J., & Beschata, R. L. (2012). Trophic Cascades in Yellowstone: The First 15
Years after Wolf Reintroduction. Biological Conservation, 145, 205-213.
https://doi.org/10.1016/j.biocon.2011.11.005
Schwarz, C. V., Reiser, B. J., Davis, E. A., Kenyon, L., Achér, A., Fortus, D., Krajcik, J. et
al. (2009). Developing a Learning Progression for Scientific Modeling: Making Scien-
tific Modeling Accessible and Meaningful for Learners. Journal of Research in Science
Teaching, 46, 632-654. https://doi.org/10.1002/tea.20311
Stroup, W. M., Ares, N., & Hurford, A. C. (2004). A Taxonomy of Generative Activity
Design Supported by Next-Generation Classroom Networks. Proceedings of the 28th
Annual Conference of Psychology in Mathematics Education-North America, Ontario,
837-846.
Stroup, W. M., Ares, N., Lesh, R., & Hurford, A. (2007). Diversity by Design: Generativity
in Next-Generation Classroom Networks. In R. Lesh, E. Hamilton, & J. J. Kaput (Eds.),
Models and Modeling as Foundations for the Future in Mathematics Education (pp.
367-393). Mahwah, NJ: Erlbaum.
Wells, G. (1993). Reevaluating the IRF Sequence: A Proposal for the Articulation of
Theories of Activity and Discourse for the Analysis of Teaching and Learning in the
Classroom. Linguistics and Education, 5, 1-37.

DOI: 10.4236/ce.2018.94043 623 Creative Education

A. J. Petrosino et al.

https://doi.org/10.1016/S0898-5898(05)80001-4
Wilensky, U. (1997). NetLogo Wolf Sheep Predation Model. Evanston, IL: Northwestern
University, Center for Connected Learning and Computer-Based Modeling.
http://ccl.northwestern.edu/netlogo/models/WolfSheepPredation
Wilensky, U., & Reisman, K. (2006). Thinking Like a Wolf, a Sheep, or a Firefly: Learning
Biology through Constructing and Testing Computational Theories—An Embodied
Modeling Approach. Cognition and Instruction, 24, 171-209.
https://doi.org/10.1207/s1532690xci2402_1
Wilensky, U., & Stroup, W. (2002). Computer-HubNet. Evanston, IL: Northwestern
University. http://ccl.northwestern.edu/netlogo/hubnet.html
Wittrock, M. C. (1991). Generative Teaching of Comprehension. The Elementary School
Journal, 92, 169-184. https://doi.org/10.1086/461686

DOI: 10.4236/ce.2018.94043 624 Creative Education